Thread: password administration
Hi postgres gurus: I would like to set up a facility that enforces password changes for roles After a predefined period (30 days for instance) when logging into psql Or, at the very least, send an email out to notify that your current Password period is about to expire. Preferably, I'd like to use The 'rolvaliduntil' column in pg_roles. I'm wondering if there is an app inside or outside of postgres that I can use or do I have to design from scratch. Thanks for your time, Mark Steben | Database Administrator @utoRevenue® - "Keeping Customers Close" 95D Ashley Ave, West Springfield, MA 01089 413.243.4800 x1512 (Phone) |413.732-1824 (Fax) @utoRevenue is a registered trademark and a division of Dominion Enterprises
On 8/5/10 12:58 PM, Mark Steben wrote: > I would like to set up a facility that enforces password changes for roles > After a predefined period (30 days for instance) when logging into psql > Or, at the very least, send an email out to notify that your current > Password period is about to expire. Preferably, I'd like to use > The 'rolvaliduntil' column in pg_roles. > > I'm wondering if there is an app inside or outside of postgres > that I can use or do I have to design from scratch. This is an off-topic response, but security experts have said that this is a REALLY bad idea. It forces people to choosea new password, which means they can't remember it. So what do they do? They write it down. Anyone snooping aroundtheir office can find it. Besides, when a password is stolen, it's usually used within minutes. Making everyone change every month does no good atall. A better solution is to implement a password-strength algorithm and require people to select decent passwords to begin with. Craig
On Thu, Aug 5, 2010 at 2:20 PM, Craig James <craig_james@emolecules.com> wrote: > On 8/5/10 12:58 PM, Mark Steben wrote: >> >> I would like to set up a facility that enforces password changes for roles >> After a predefined period (30 days for instance) when logging into psql >> Or, at the very least, send an email out to notify that your current >> Password period is about to expire. Preferably, I'd like to use >> The 'rolvaliduntil' column in pg_roles. >> >> I'm wondering if there is an app inside or outside of postgres >> that I can use or do I have to design from scratch. > > This is an off-topic response, but security experts have said that this is a > REALLY bad idea. It forces people to choose a new password, which means > they can't remember it. So what do they do? They write it down. Anyone > snooping around their office can find it. > > Besides, when a password is stolen, it's usually used within minutes. > Making everyone change every month does no good at all. > > A better solution is to implement a password-strength algorithm and require > people to select decent passwords to begin with. Exactly. If you allow simpler passwords that have to be changed you get things like: ilovemywife22 md5: b845aec254d018d118fe52c46ee8c98c changed to ilovemywife23 md5: 8c2b59e4d961478e3a9d5bd94979f329 You can't tell how close they are by the md5. If you try to prevent people from reusing similar passwords, then you have to store either the previous passwords (bad security) or something like a soundex of the previous password (also bad security.) Better to have strong passwords and intrusion detection.
"Mark Steben" <msteben@autorevenue.com> writes:
> I would like to set up a facility that enforces password changes for roles
> After a predefined period (30 days for instance) when logging into psql
> Or, at the very least, send an email out to notify that your current
> Password period is about to expire.
Usually we suggest using PAM when you want to do this, as there's all
sorts of spare parts out there already for PAM-managed passwords.
(I concur with the response questioning whether forced password changes
are good policy, especially with an interval as short as that. But if
you've got bullheaded management to deal with, PAM is the place to
look.)
regards, tom lane
On Thu, Aug 5, 2010 at 2:27 PM, Scott Marlowe <scott.marlowe@gmail.com> wrote: > On Thu, Aug 5, 2010 at 2:20 PM, Craig James <craig_james@emolecules.com> wrote: >> On 8/5/10 12:58 PM, Mark Steben wrote: >>> >>> I would like to set up a facility that enforces password changes for roles >>> After a predefined period (30 days for instance) when logging into psql >>> Or, at the very least, send an email out to notify that your current >>> Password period is about to expire. Preferably, I'd like to use >>> The 'rolvaliduntil' column in pg_roles. >>> >>> I'm wondering if there is an app inside or outside of postgres >>> that I can use or do I have to design from scratch. >> >> This is an off-topic response, but security experts have said that this is a >> REALLY bad idea. It forces people to choose a new password, which means >> they can't remember it. So what do they do? They write it down. Anyone >> snooping around their office can find it. >> >> Besides, when a password is stolen, it's usually used within minutes. >> Making everyone change every month does no good at all. >> >> A better solution is to implement a password-strength algorithm and require >> people to select decent passwords to begin with. > > Exactly. If you allow simpler passwords that have to be changed you > get things like: > > ilovemywife22 md5: b845aec254d018d118fe52c46ee8c98c > > changed to > > ilovemywife23 md5: 8c2b59e4d961478e3a9d5bd94979f329 > > You can't tell how close they are by the md5. If you try to prevent > people from reusing similar passwords, then you have to store either > the previous passwords (bad security) or something like a soundex of > the previous password (also bad security.) > > Better to have strong passwords and intrusion detection. Also, if you do get strong passwords but then force people to change them every 30 days, they wind up forgetting them. Which leads to two other unintended problems. Either the admins get real used to resetting passwords and get lazy about checking who they're resetting passwords for and when, creating openings for social engineering hacks, or people start writing down their complex passwords and putting them under their keyboards etc. If you require annual or bi-annual password changes you're far less likely to have these issues pop up. Take it to the extreme, how about requiring password changes every hour, or 4 times a day? How would you methods now change? Something similar to this is those key fobs that generate a new pseudo random key every x seconds, and you have to put in a recent one to log in. That is far more uesful for security than changing passwords often is. Anyway, like Tom said, externalize it with PAM.
Scott Marlowe <scott.marlowe@gmail.com> writes:
> On Thu, Aug 5, 2010 at 2:20 PM, Craig James <craig_james@emolecules.com> wrote:
>> A better solution is to implement a password-strength algorithm and require
>> people to select decent passwords to begin with.
> Exactly. If you allow simpler passwords that have to be changed you
> get things like:
> ilovemywife22 md5: b845aec254d018d118fe52c46ee8c98c
> changed to
> ilovemywife23 md5: 8c2b59e4d961478e3a9d5bd94979f329
> You can't tell how close they are by the md5. If you try to prevent
> people from reusing similar passwords, then you have to store either
> the previous passwords (bad security) or something like a soundex of
> the previous password (also bad security.)
A place I know but won't name has a policy of storing your last five
passwords (hopefully in md5'd form, but I don't actually know that) and
not letting you reuse those. Of course this merely encourages people to
use a cycle of six or so passwords, like something they can remember
with one digit tagged on.
The real problem with any such policy is that what you actually want is
for people to use strong passwords that they can remember, and they are
never going to remember a good password if they have to invent a new one
every month. Aforesaid place has an administrative system that I have
to use maybe two or three times a year, with a password change interval
that is shorter than my average time between uses. Am I going to invent
a strong password that I will need to change the very next time I have
use for it? If I take the trouble, will I actually remember it next
time? Not a chance. So I write it down ... and I'm way more security
conscious than most people. Their loss for having a brain-dead password
policy.
regards, tom lane
Also, for reference: http://research.microsoft.com/en-us/people/cormac/ Specifically: http://research.microsoft.com/users/cormac/papers/2009/SoLongAndNoThanks.pdf
Mark, You could use LDAP to? that? I use LDAP + apache []´s Victor Hugo 2010/8/5 Tom Lane <tgl@sss.pgh.pa.us>: > "Mark Steben" <msteben@autorevenue.com> writes: >> I would like to set up a facility that enforces password changes for roles >> After a predefined period (30 days for instance) when logging into psql >> Or, at the very least, send an email out to notify that your current >> Password period is about to expire. > > Usually we suggest using PAM when you want to do this, as there's all > sorts of spare parts out there already for PAM-managed passwords. > > (I concur with the response questioning whether forced password changes > are good policy, especially with an interval as short as that. But if > you've got bullheaded management to deal with, PAM is the place to > look.) > > regards, tom lane > > -- > Sent via pgsql-admin mailing list (pgsql-admin@postgresql.org) > To make changes to your subscription: > http://www.postgresql.org/mailpref/pgsql-admin > -- []´s Victor Hugo
Correcting.... I use LDAP + postgresql 2010/8/5 Victor Hugo <vh.clemente@gmail.com>: > Mark, > > You could use LDAP to? that? > > I use LDAP + apache > > []´s > Victor Hugo > > > > 2010/8/5 Tom Lane <tgl@sss.pgh.pa.us>: >> "Mark Steben" <msteben@autorevenue.com> writes: >>> I would like to set up a facility that enforces password changes for roles >>> After a predefined period (30 days for instance) when logging into psql >>> Or, at the very least, send an email out to notify that your current >>> Password period is about to expire. >> >> Usually we suggest using PAM when you want to do this, as there's all >> sorts of spare parts out there already for PAM-managed passwords. >> >> (I concur with the response questioning whether forced password changes >> are good policy, especially with an interval as short as that. But if >> you've got bullheaded management to deal with, PAM is the place to >> look.) >> >> regards, tom lane >> >> -- >> Sent via pgsql-admin mailing list (pgsql-admin@postgresql.org) >> To make changes to your subscription: >> http://www.postgresql.org/mailpref/pgsql-admin >> > > > > -- > []´s > Victor Hugo > -- []´s Victor Hugo
Mark, Look into kerberos. You will have to recompile your server to use it. Bob Lunney --- On Thu, 8/5/10, Mark Steben <msteben@autorevenue.com> wrote: > From: Mark Steben <msteben@autorevenue.com> > Subject: [ADMIN] password administration > To: pgsql-admin@postgresql.org > Date: Thursday, August 5, 2010, 3:58 PM > > > Hi postgres gurus: > > I would like to set up a facility that enforces password > changes for roles > After a predefined period (30 days for instance) when > logging into psql > Or, at the very least, send an email out to notify that > your current > Password period is about to expire. Preferably, I'd > like to use > The 'rolvaliduntil' column in pg_roles. > > I'm wondering if there is an app inside or outside of > postgres > that I can use or do I have to design from scratch. > > Thanks for your time, > > > Mark Steben | Database Administrator > @utoRevenue® - "Keeping Customers Close" > 95D Ashley Ave, West Springfield, MA 01089 > 413.243.4800 x1512 (Phone) |413.732-1824 (Fax) > @utoRevenue is a registered trademark and a division of > Dominion > Enterprises > > > > > > -- > Sent via pgsql-admin mailing list (pgsql-admin@postgresql.org) > To make changes to your subscription: > http://www.postgresql.org/mailpref/pgsql-admin >
On 05.08.2010 22:54, Tom Lane wrote: > Scott Marlowe <scott.marlowe@gmail.com> writes: > >> On Thu, Aug 5, 2010 at 2:20 PM, Craig James <craig_james@emolecules.com> wrote: >> >>> A better solution is to implement a password-strength algorithm and require >>> people to select decent passwords to begin with. >>> > >> Exactly. If you allow simpler passwords that have to be changed you >> get things like: >> > >> ilovemywife22 md5: b845aec254d018d118fe52c46ee8c98c >> > >> changed to >> > >> ilovemywife23 md5: 8c2b59e4d961478e3a9d5bd94979f329 >> > >> You can't tell how close they are by the md5. If you try to prevent >> people from reusing similar passwords, then you have to store either >> the previous passwords (bad security) or something like a soundex of >> the previous password (also bad security.) >> > A place I know but won't name has a policy of storing your last five > passwords (hopefully in md5'd form, but I don't actually know that) and > not letting you reuse those. Of course this merely encourages people to > use a cycle of six or so passwords, like something they can remember > with one digit tagged on. > Hi! Such a policy is in force in my country (Poland) but only if system contains personal data. (government law) 8 or more characters - 2 capital letters, 2 digits And... sometimes this is pain in the... but we don't have a choice. TIP: you don't need 6 passwords - just 2 - with different one character ;-) -- Andrzej Zawadzki